Monitoring temps réel

Collected metrics

The Aspirateur continuously collects over 120 metrics on each MariaDB / MySQL server:

• Performance — Queries/sec, slow queries, threads running/connected, table locks
• InnoDB — Buffer pool hit ratio, dirty pages, redo log, row locks, deadlocks
• Replication — Seconds behind master, IO/SQL thread status, GTID
• Connections — Active connections, max_connections, aborted connects/clients
• Cache — Query cache hit ratio, key buffer usage, table open cache
• Disk — Space used by datadir, binlogs, tmp, logs
• Hardware — CPU, RAM, swap, load average (via SSH)

Dashboard

The Dashboard (Dashboard.php) aggregates key metrics:

• Hit Ratio — InnoDB buffer pool and query cache efficiency
• Thread Cache — Thread reuse ratio
• Open Files — Open files vs limit
• Lock Ratio — Table lock vs row-level lock ratio

Data is exposed as JSON for integration with external dashboards (Grafana, etc.).

Server comparison

The Monitoring module allows comparing metrics between two or more servers. Useful for:

• Validating that a replica has the same performance as the master
• Comparing configurations between environments
• Detecting divergences after a failover

Post-mortem

The PostMortem module (PostMortem.php) allows analyzing system metrics after an incident. It formats performance data for detailed chronological analysis.